Btrfs: add skeleton code for compression heuristic

Add skeleton code for compresison heuristics. Now it iterates over all
the pages, but in the end always says "yes, compress please", ie it does
not change the current behaviour.

In the future we're going to add various heuristics to analyze the data.
This patch can be used as a baseline for measuring if the effectivness
and performance.

Signed-off-by: Timofey Titovets <nefelim4ag@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ enhanced changelog, modified comments ]
Signed-off-by: David Sterba <dsterba@suse.com>

diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 3896bd0175ecaa27ec961d95c44ce3dea0426aa2..883ecc58fd0dabeda45dd05557c503bd47a71902 100644 (file)
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -1047,3 +1047,36 @@ int btrfs_decompress_buf2page(const char *buf, unsigned long buf_start,
 
        return 1;
 }
+
+/*
+ * Compression heuristic.
+ *
+ * For now is's a naive and optimistic 'return true', we'll extend the logic to
+ * quickly (compared to direct compression) detect data characteristics
+ * (compressible/uncompressible) to avoid wasting CPU time on uncompressible
+ * data.
+ *
+ * The following types of analysis can be performed:
+ * - detect mostly zero data
+ * - detect data with low "byte set" size (text, etc)
+ * - detect data with low/high "core byte" set
+ *
+ * Return non-zero if the compression should be done, 0 otherwise.
+ */
+int btrfs_compress_heuristic(struct inode *inode, u64 start, u64 end)
+{
+       u64 index = start >> PAGE_SHIFT;
+       u64 end_index = end >> PAGE_SHIFT;
+       struct page *page;
+       int ret = 1;
+
+       while (index <= end_index) {
+               page = find_get_page(inode->i_mapping, index);
+               kmap(page);
+               kunmap(page);
+               put_page(page);
+               index++;
+       }
+
+       return ret;
+}

diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index 87f6d3332163e2949465a9b85f8c7c73d963e584..8508ba6b9aef6deead0c534a3dced743fa69b8df 100644 (file)
--- a/fs/btrfs/compression.h
+++ b/fs/btrfs/compression.h
@@ -129,4 +129,6 @@ struct btrfs_compress_op {
 extern const struct btrfs_compress_op btrfs_zlib_compress;
 extern const struct btrfs_compress_op btrfs_lzo_compress;
 
+int btrfs_compress_heuristic(struct inode *inode, u64 start, u64 end);
+
 #endif

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 467b9477dac4d9cc34bdd4094cf5447aeaa9f775..0bd008b9e0d162325fea7434ceca51ec9893791d 100644 (file)
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -392,7 +392,7 @@ static noinline int add_async_extent(struct async_cow *cow,
        return 0;
 }
 
-static inline int inode_need_compress(struct inode *inode)
+static inline int inode_need_compress(struct inode *inode, u64 start, u64 end)
 {
        struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 
@@ -405,7 +405,7 @@ static inline int inode_need_compress(struct inode *inode)
        if (btrfs_test_opt(fs_info, COMPRESS) ||
            BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS ||
            BTRFS_I(inode)->force_compress)
-               return 1;
+               return btrfs_compress_heuristic(inode, start, end);
        return 0;
 }
 
@@ -503,7 +503,7 @@ again:
         * inode has not been flagged as nocompress.  This flag can
         * change at any time if we discover bad compression ratios.
         */
-       if (inode_need_compress(inode)) {
+       if (inode_need_compress(inode, start, end)) {
                WARN_ON(pages);
                pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
                if (!pages) {
@@ -1576,7 +1576,7 @@ static int run_delalloc_range(void *private_data, struct page *locked_page,
        } else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC && !force_cow) {
                ret = run_delalloc_nocow(inode, locked_page, start, end,
                                         page_started, 0, nr_written);
-       } else if (!inode_need_compress(inode)) {
+       } else if (!inode_need_compress(inode, start, end)) {
                ret = cow_file_range(inode, locked_page, start, end, end,
                                      page_started, nr_written, 1, NULL);
        } else {